Novi, a platform helping brands optimise product data for AI shopping assistants, has filed a provisional patent application with the US Patent and Trademark Office. The filing covers a proprietary methodology that extracts and maps how large language models select and rank products within a category. The patent-pending system underpins Novi's AI Intelligence product, which is based on the insight that large language models organise meaning around semantic neighbourhoods—clusters of related concepts capturing various consumer phrasings. The methodology reveals how LLMs research and select products, enabling brands to influence when their products appear regardless of prompt wording. Founded in 2020 and backed by Tiger Global, Defy and Greylock, Novi serves as a bridge between brands, certification organisations and major retailers including Target, Ulta and Sephora.

Novi announced its new Novi AI Shopping Optimizer, designed to increase product sales by helping brands and retailers surface products backed by the verified, trustworthy information that AI shopping assistants prioritize.